Detection and parameter estimation of multiple nonGaussian sources via higher order statistics

Simultaneous detection of signal arriving at a sensor array and estimation of their parameters is carried out using higher than second-order statistics. Information theoretic criteria which are (at least theoretically) insensitive to additive Gaussian noise are developed to estimate consistently the parameters as well as the number of non-Gaussian but unknown sources. The novel cumulant based algorithms can estimate parameters of more sources with fewer sensors. Simulations confirm superior resolution capability of the proposed methods for both narrow-band and wideband sources in the presence of low SNR additive correlated Gaussian noise